Bicycles are among the simplest and at the same time most fascinating vehicles. Despite their apparently basic design, their stability, dynamics, performance, and safety are still largely unknown. In a world where cities suffer from critical air pollution levels and are often congested by car traffic, bicycles can represent a cost-effective and ready solution to address part of the problem. However, the increasing use of bicycles enlarges the risk of bicycle-related accidents and injuries. Many of these accidents involve falls, highlighting the need of proper studies on bicycle-rider dynamics. While research has been conducted on the effect of bicycle parameters on the handling quality properties, we still do not know enough about the specific role of tyres on bicycle-rider dynamics. To better simulate what happens in the real world, existing bicycle mathematical models need to be integrated with robust tyre models that include side slip, as they can change both the lateral and longitudinal dynamics. Actually, the use of nonlinear tyre models allows taking into account the lateral force and self-aligning torque saturation. This is needed for instance to model the wobble as a non-divergent vibration modes, contrarily to what happens when using a simple linear tyre model. These features relate directly to predict critical situations for which we need a deep knowledge of bicycle tyres, thus proper test-rigs to measure their mechanical characteristics...@en

In this research a dynamic time and energy model is constructed to simulate free flow utility cycling. Using a Monte Carlo simulation and distributions of rider and bicycle characteristics a population of cyclists is modelled to find the difference in time and energy expenditure on six different bicycle routes. The riders use realistic like power inputs and braking behaviour based on the rider and the route characteristics. While travel times are almost always lower for all cyclists when comparing routes, there are cyclists for who a shorter route distance does not equal a lower energy expenditure. For simpler routes (less elevation difference, traffic lights and shorter distance) the standard deviation of both time and energy decreases, showing that slow cyclist have a relative higher gain. For e-bike users there is even almost no difference in energy expenditure between the six evaluated routes that have varying elevation, traffic signals and routing, while travel times show a similar trend as for regular bicycles. With this model difference in routes can be quantified in matters of time and energy expenditure for a population of cyclists giving an objective picture of the differences between routes, which can be a useful tool for city planning and evaluation of bicycle infrastructure.

The rapid adoption of e-bikes as an alternative mode of transportation to automobiles gives rise to new methods of safety regulations for cyclists. Modern e-bikes feature Internet of Things (IoT) modules capable of collecting and sending cycling data that can be used for traffic safety analysis. This study explores the potential of using cycling data to detect dangerous intersections and then implement local low-speed areas using geofencing.

A natural cycling experiment with 10 participants is conducted, where data is collected similar to the capabilities of e-bike IoT modules (GNSS, power and IMU data) and participants are asked to cycle through Delft, The Netherlands. The data is combined with a traffic accident dataset from the Dutch government, where intersections with accidents are labelled dangerous. The cycling data is separated into separate intersection approaches, based on positional data. Metrics based on power and cadence data show the most significant statistical difference (p-values <0.02) and the largest effect size (Cohen's d >0.80) at dangerous intersections. A binary classification model is trained on the dataset, which can correctly predict whether an intersection is dangerous or safe with an accuracy of 68,2%, a specificity of 50,0%, and a sensitivity of 24,2%. This prediction is based on the metrics of a single intersection approach.

To investigate the viability of using geofencing to implement low-speed areas at designated dangerous intersections, a simulation is carried out to determine the minimum size of a geofence that effectively slows down e-bike cyclists. The minimal effective geofence size to limit a cyclist's speed at an intersection is determined by simulating cyclists approaching a geofence for several geofence perimeters, incline levels, and wind speeds. The e-bike's motor stops supporting when the geofence is entered and the cyclist is cycling above a target speed. The minimal effective geofence size ranges from 50 to 300+ m and depends on the target speed, wind speed, and road incline.

This study shows that cycling data can be used to identify dangerous intersections. Enforcing a slower speed on intersections with local geofencing is not feasible, as the geofences have to be extremely large, showing a lot of overlap on other intersections in cities due to high intersection density. Future work can be done on the intersection risk estimation method and the feasibility of speed-limiting geofences that employ active braking.

Cycling is an increasingly attractive transportation mode, thanks to its health and environmental benefits. Personalized travel assistance services can help make cycling more appealing by providing speed or route advices that can reduce travel time and increase safety while taking into account the personal preferences of cyclists. Due to its ability to learn agents' reward function, Inverse Reinforcement Learning is a suitable algorithm for learning cycling preferences from data.
This thesis aims to describe cycling styles as a set of cycling preferences encoded as a reward function composed of a weighted sum of features. The weights associated to the features composing the reward function represent the importance given to each cycling preference and express the trade-off between different goals of a cyclist. Continuous-time Inverse Reinforcement Learning extracts the weights from empirical cyclists' trajectories collected during an experiment performed in Delft. During the experiment, cyclists were asked to cycle according to three different cycling styles: cautious, normal and aggressive. Differences between weight sets extracted for each cycling styles were analyzed by means of the Kruskar-Wallis statistical test and K-Means clustering algorithm, and the averaged weights for each cycling style were used to simulate a set of test trajectories.
It is shown by simulations that the reward function identified for a specific cycling style leads to an improvement in terms of similarity to test trajectories with the same cycling style with respect to the reward functions corresponding to other cycling styles. The statistical analysis shows that the weights of cautious and aggressive cycling styles show statistical differences and define separate clusters.

Skateboarding involves a human controlling a four wheeled vehicle that is steered by tilting the standing surface. The riding mechanics of skateboarding have been well reported [2, 3]. The sport also includes aerial maneuvers such as jumping of stairs, flying off ramps and flipping and rotating the skateboard. The most basic aerial trick is called the ollie. The athlete jumps up while pushing down on the back end of the skateboard’s tail, causing a rotation about the back axle. The upward acceleration due to the rotation together with the tail-ground impact cause the skateboard to go airborne. Midair the athlete drags the skateboard up through frictional contact and levels it out to land the trick. The most concrete performance measure of the ollie is height according to the Olympic judging criteria [4]. To reach maximum height the dynamics such as impact, dynamic response, and torque production are dependent on shape, inertia and mass, which gives reason to assume an optimal shape exists. This leads to the research question: What are the optimal geometric and inertial parameters of a skateboard for an Olympic athlete to reach maximal ollie height. The skateboard geometry is optimized through multiphase direct collocation with the objective of maximal ollie height. A parameterized model is created with scaling mass and inertia properties such that the geometry of the skateboard. Modelling the dynamics of the ollie including impact and friction are done with a point mass human controller that is kineticly and kinematicly mapped to a counter movement jump. A simplistic contact implicit impact scheme is made for a higher order optimization. The ollie height is improved by changing the mass and inertia properties of the skateboard. Multiple optimal board shapes are generated for example a skateboard with a smaller wheelbase can reach higher ollie height compared to an industry standard skateboard.

Bicycles connected to the internet present an opportunity for integrated accident detection and geolocation. Such a system can reduce the time it takes for help to arrive by automatically alerting predefined contacts with the location of the accident. I developed a systematic method for the practical implementation of bicycle accident detection in connected bicycles and present the performance of a prototype system. The method uses accelerometer and gyroscopic measurements as well as localization and velocity estimations. Supplementing existing research, a bicycle accident detection system is validated on normal cycling, edge cases, and three types of single bicycle accidents with constraints set by a bicycle manufacturer. Edge cases are movements of a bicycle that occur during regular usage, but can not be described by normal cycling. This method uses a data¬driven approach. For the prototype system, the input signals are collected during 71 different simulated accidents and 54 hours of normal cycling and edge cases. A three¬layer detection algorithm determines if an accident has occurred and sends the last known location to a set of predefined contacts. Multiple combinations of thresholds and classification algorithms are compared. This resulted in a prototype system with a K¬Nearest Neighbours classifier which detects 75% of accidents. Normal cycling and edge cases are correctly detected 99.997% of the time. From all warnings send, 85.7% are true accidents. The prototype system proves that the proposed method can be used to integrate reliable accident detection in connected bicycles. Bicycles with such a system automatically inform emergency contacts with a message containing the location of the accident, in a time where every second counts.